1. Field of the Invention
The present disclosure relates to obtaining artist and title information via an acoustic recognition system for music transmitted on broadcast radio.
2. Description of the Prior Art
In the radio broadcast industry, the over-the-air broadcast is referred to as a “terrestrial” broadcast, differentiating from satellite and Internet broadcasts. The delivery of a terrestrial broadcast on the Internet is commonly referred to as webcasting, simulcasting, or streaming. Beyond commercial AM/FM radio broadcasts that simulcast on the Internet, there are numerous Internet-only music broadcasters, such as Pandora, Beethoven, last.fm, iTunes, and many others, whose listenership has become a very significant portion of all Internet music listenership. In addition to these sources of music, there are other technologies of interest, as follows.
Broadcast Radio Provisioning of Artist and Title for Songs
As of this writing in 2012, there are two established broadcast radio technologies that enable a radio station to transmit artist and title metadata associated with the currently playing song for presentation on a display associated with the radio receiver. These two technologies are known as RDS, and HD radio. For either of these technologies to work the radio station must have a mechanism for associating metadata with a song. The mainstream approach is the use of a Broadcast Automation System, which holds the digitized versions of the music along with the metadata for that music, and incorporates a scheduling system which actually plays out the music and sends it to the radio station's audio chain which delivers it to the transmitter system. The Broadcast Automation System also has a digital data output which can output the metadata associated with a song, or other data, to an external system. In the case of RDS, for example, that metadata will be sent to an RDS digital encoder, typically a standalone box, which will modulate the textual metadata onto an FM subcarrier; the subcarrier then gets mixed with, and transmitted with, the broadcast audio signal. Radio receivers equipped for RDS will decode the RDS data stream for the tuned-in station, extract the artist and title information, and cause the display that information in an appropriate place or manner on an associated display screen. In a case of auto radio receivers, that display screen might be part of the radio itself, or it might be a multipurpose display that is shared with vehicle navigation and other vehicle functionality. For both RDS and HD radio, it has been a common practice in the industry for the automation system to send radio station identifiers such as the radio station call letters, frequency, slogan, and nickname (“The Power”, “Kiss 107”, “The Breeze”) as part of the metadata output.
Digital Signal Processing, and Beat Detection
There is an associated and well-known technology, that of beat detection, identification and beat matching in systems used by DJs and radio stations to segue from one song to the next. Beat matching is a well-studied mathematical science in the field of digital signal processing, and there are commonly available software modules that a programmer can incorporate into his system to build in beat identification and beat matching.
Acoustic Fingerprinting of Audio Materials and Music
Further, there are systems and service providers on the Internet that utilize a technology called acoustic fingerprinting (or audio recognition, acoustic recognition, or similar terms) which uses a mathematical algorithm to associate a digital “fingerprint” of audio material, such as a song or commercial, with the name or other identifier of the audio material itself from an audio content database (a song database). A song database is comprised of records of songs wherein each record is a mathematical representation of a song, typically consisting of a series of fingerprints of various points in the song. Each record is associated with one or more identifiers of the song, particularly the artist and title of the song. In one method of operation of these systems for the recognition of songs, a snippet of a song is sent to the system in a digitized format such as MP3, the snippet is fingerprinted, the fingerprint is compared against a database of fingerprints, and if a close match is found, that match identifies the song from which the snippet is derived. Having identified the song, the system then returns information to the requester about the song, such as the artist and title and album information, or takes other appropriate actions according to the business nature of the service provider. While there may be intermediate steps on the part of any individual system or service provider, from the viewpoint of that system or service as a utility the operation is the same: pass in a snippet of a song, and receive back metadata about the song if it exists in their database.
According to common knowledge, acoustic fingerprinting was developed by the US military decades ago to recognize voices. Although esoteric, it is well-known and well-studied in certain mathematical fields, especially those associated with digital signal processing. As of this writing there are multiple services available on the Internet that provide song identification via acoustic fingerprinting in a fashion comparable to that just described.
In one implementation, acoustic fingerprinting service providers (such as Media Monitors LLC, ASCAP [American Society of Composers, Authors and Publishers], and Broadcast Music Inc. [BMI]) provide a monitoring and auditing function for radio stations, or other clients that wish to audit radio stations, to verify that a radio station has played their commercials according to agreement, or to identify and track the songs played by a radio station. In this implementation, the service provider will arrange to have one or more radio receivers installed a listening area, one receiver per radio station wherein the receiver is permanently tuned to that station. The audio output (line out, or headphone out) of each receiver is physically connected to a co-located computer. A computer may be provisioned with multiple sound cards to allow for connections from multiple receivers (i.e., four sound cards will allow the connection of four radio receivers). Thus, to monitor 1,000 radio stations in 60 listening areas, the service provider must arrange for the installation and maintenance of 1,000 radio receivers in 60 physical locations. If the associated computers were configured with four sound cards each (assuming there is no other limiting parameter), 250 computers are required. The computers digitize the audio from each associated radio receiver into a suitable format such as mp3 and send the stream of digitized audio to the service provider for acoustic fingerprinting. The service provider analyzes each received audio stream using fingerprinting techniques to determine the commercials played and time-of-day, which information is subsequently made available in reporting formats to the clients of the service provider.
Some of the agencies listed above, plus others such as TuneSat.com, provide similar services on behalf of music rights holders by monitoring television channels to detect music embedded in TV programs or commercials.
In a variation of this business model, the acoustic fingerprinting service providers (such as Yes.com, Nielsen), configured as just described, determine the artist and title of songs that have played, and make that information available to their radio station clients via proprietary Internet connection arrangements in a controlled and limited fashion so song information can be displayed on the radio station's website, or on an Internet media player configured to play the radio station's content. In these configurations, the use of a proprietary connection enables the display of only the most current song, and induces the audience to click through to the service provider's own website to access the listings of previous songs. The service providers monetize the visits of listeners that click through in this fashion by the presentation of ads, and by offering songs on the playlist for sale.
In another quite different implementation, acoustic fingerprinting service providers (such as Shazam and SoundHound) provide a smart phone application for users. In this implementation, when a user hears an interesting song from an ambient source (background music in a store or office, a friend's radio, etc.), the user turns on their smart phone and starts the app. The app turns on the microphone of the smart phone, digitizes the audio it “hears”, and sends that audio stream to the service provider. The service provider fingerprints the audio, as previously described, determines the name of the song and the artist, and delivers that information back to the app where it is displayed for the user. The song is then offered for sale, and the user is presented opportunities to explore similar songs, and to purchase them as well.
Device that Hears What You Hear
Arbitron provides audience measurement and ratings for the radio broadcast industry in the United States. Traditionally Arbitron enlisted volunteers to self-report their listening history by recording that information in a diary and mailing it back to Arbitron. Multiple volunteers are enlisted in each listening area targeted for study. A few years ago Arbitron introduced an automated technology called the Portable People Meter (PPM), a wearable device with a continuously-on microphone that hears what the wearer hears. Radio stations participating in Arbitron surveys install an encoder that impresses a unique identifying code in the form of an inaudible series of tones onto the radio station's broadcast signal. PPMs listen for these inaudible tones, and record the number of instances, duration, and time of day for each station the volunteer has listened to during the study period. Only radio stations transmitting these codes can be detected by the PPM. Although the radio station audio is heard by the PPM in order to detect the tones, the nature or content of the audio other than the tones is ignored by the PPM and is neither identified nor recorded. The PPMs are inserted into a dock at night which charges the PPM and uploads the data to Arbitron servers.